INDEX

Explanations

words related to responses or replies in the context of conversations

oai_token-act-pair · gpt-3.5-turbo

instances of dialogue or spoken responses

oai_token-act-pair · gpt-4o-mini Triggered by @bot

New Auto-Interp

Top Features by Cosine Similarity

Comparing With GPT2-SMALL @ 4-res-jb

Configuration

jbloom/GPT2-Small-SAEs-Reformatted/blocks.4.hook_resid_pre

Prompts (Dashboard)

24,576 prompts, 128 tokens each

Dataset (Dashboard)

Skylion007/openwebtext

Features

24,576

Data Type

torch.float32

Hook Point

blocks.4.hook_resid_pre

Architecture

standard

Context Size

128

Dataset

Skylion007/openwebtext

Hook Point Layer

Activation Function

relu

Embeds

IFrame

Link

Not in Any Lists

No Comments

Negative Logits

ctors

-0.77

olin

-0.73

cipled

-0.71

bons

-0.71

icipated

-0.69

dar

-0.68

wed

-0.68

prus

-0.67

gone

-0.65

ciples

-0.64

POSITIVE LOGITS

 angrily

0.88

 sarcast

0.88

 thereto

0.83

 favorably

0.82

 affirm

0.80

reply

0.77

 indign

0.73

 harshly

0.72

 whine

0.72

 replies

0.71

Activations Density 0.028%

words related to responses or replies in the context of conversations

instances of dialogue or spoken responses

No Comments

No Known Activations